Examples with SmooshedFileMapper org.apache.druid.java.util.common.io.smoosh.SmooshedFileMapper used on opensource projects

Example 1 with SmooshedFileMapper

use of org.apache.druid.java.util.common.io.smoosh.SmooshedFileMapper in project druid by druid-io.

the class V3CompressedVSizeColumnarMultiIntsSerializerTest method generateV2SerializedSizeAndData.

private void generateV2SerializedSizeAndData(long numRows, int maxValue, int maxValuesPerRow, int offsetChunkFactor, int valueChunkFactor) throws Exception {
    File tmpDirectory = FileUtils.createTempDir(StringUtils.format("CompressedVSizeIndexedV3WriterTest_%d_%d", offsetChunkFactor, offsetChunkFactor));
    FileSmoosher smoosher = new FileSmoosher(tmpDirectory);
    try (SegmentWriteOutMedium segmentWriteOutMedium = TmpFileSegmentWriteOutMediumFactory.instance().makeSegmentWriteOutMedium(temporaryFolder.newFolder())) {
        CompressedColumnarIntsSerializer offsetWriter = new CompressedColumnarIntsSerializer(TEST_COLUMN_NAME, segmentWriteOutMedium, offsetChunkFactor, byteOrder, compressionStrategy, GenericIndexedWriter.ofCompressedByteBuffers(segmentWriteOutMedium, "offset", compressionStrategy, Long.BYTES * 250000));
        GenericIndexedWriter genericIndexed = GenericIndexedWriter.ofCompressedByteBuffers(segmentWriteOutMedium, "value", compressionStrategy, Long.BYTES * 250000);
        CompressedVSizeColumnarIntsSerializer valueWriter = new CompressedVSizeColumnarIntsSerializer(TEST_COLUMN_NAME, segmentWriteOutMedium, maxValue, valueChunkFactor, byteOrder, compressionStrategy, genericIndexed);
        V3CompressedVSizeColumnarMultiIntsSerializer writer = new V3CompressedVSizeColumnarMultiIntsSerializer(TEST_COLUMN_NAME, offsetWriter, valueWriter);
        writer.open();
        for (long l = 0L; l < numRows; l++) {
            writer.addValues(new ArrayBasedIndexedInts(generateRow(rand, maxValue, maxValuesPerRow)));
        }
        final SmooshedWriter channel = smoosher.addWithSmooshedWriter("test", writer.getSerializedSize());
        writer.writeTo(channel, smoosher);
        channel.close();
        smoosher.close();
        SmooshedFileMapper mapper = Smoosh.map(tmpDirectory);
        V3CompressedVSizeColumnarMultiIntsSupplier supplierFromByteBuffer = V3CompressedVSizeColumnarMultiIntsSupplier.fromByteBuffer(mapper.mapFile("test"), byteOrder, mapper);
        ColumnarMultiInts columnarMultiInts = supplierFromByteBuffer.get();
        Assert.assertEquals(columnarMultiInts.size(), numRows);
        Random verifier = new Random(0);
        for (int i = 0; i < numRows; ++i) {
            IndexedInts subVals = columnarMultiInts.get(i);
            int[] expected = generateRow(verifier, maxValue, maxValuesPerRow);
            Assert.assertEquals(subVals.size(), expected.length);
            for (int j = 0, size = subVals.size(); j < size; ++j) {
                Assert.assertEquals(subVals.get(j), expected[j]);
            }
        }
        CloseableUtils.closeAll(columnarMultiInts, mapper);
    }
}

Example 2 with SmooshedFileMapper

use of org.apache.druid.java.util.common.io.smoosh.SmooshedFileMapper in project druid by druid-io.

the class LargeColumnSupportedComplexColumnSerializerTest method testSanity.

@Test
public void testSanity() throws IOException {
    HyperUniquesSerdeForTest serde = new HyperUniquesSerdeForTest(Hashing.murmur3_128());
    int[] cases = { 1000, 5000, 10000, 20000 };
    int[] columnSizes = { Integer.MAX_VALUE, Integer.MAX_VALUE / 2, Integer.MAX_VALUE / 4, 5000 * Long.BYTES, 2500 * Long.BYTES };
    for (int columnSize : columnSizes) {
        for (int aCase : cases) {
            File tmpFile = temporaryFolder.newFolder();
            HyperLogLogCollector baseCollector = HyperLogLogCollector.makeLatestCollector();
            try (SegmentWriteOutMedium segmentWriteOutMedium = new OffHeapMemorySegmentWriteOutMedium();
                FileSmoosher v9Smoosher = new FileSmoosher(tmpFile)) {
                LargeColumnSupportedComplexColumnSerializer serializer = LargeColumnSupportedComplexColumnSerializer.createWithColumnSize(segmentWriteOutMedium, "test", serde.getObjectStrategy(), columnSize);
                serializer.open();
                for (int i = 0; i < aCase; i++) {
                    HyperLogLogCollector collector = HyperLogLogCollector.makeLatestCollector();
                    byte[] hashBytes = fn.hashLong(i).asBytes();
                    collector.add(hashBytes);
                    baseCollector.fold(collector);
                    serializer.serialize(new ObjectColumnSelector() {

                        @Nullable
                        @Override
                        public Object getObject() {
                            return collector;
                        }

                        @Override
                        public Class classOfObject() {
                            return HyperLogLogCollector.class;
                        }

                        @Override
                        public void inspectRuntimeShape(RuntimeShapeInspector inspector) {
                        // doesn't matter in tests
                        }
                    });
                }
                try (final SmooshedWriter channel = v9Smoosher.addWithSmooshedWriter("test", serializer.getSerializedSize())) {
                    serializer.writeTo(channel, v9Smoosher);
                }
            }
            SmooshedFileMapper mapper = Smoosh.map(tmpFile);
            final ColumnBuilder builder = new ColumnBuilder().setType(ValueType.COMPLEX).setHasMultipleValues(false).setFileMapper(mapper);
            serde.deserializeColumn(mapper.mapFile("test"), builder, null);
            ColumnHolder columnHolder = builder.build();
            ComplexColumn complexColumn = (ComplexColumn) columnHolder.getColumn();
            HyperLogLogCollector collector = HyperLogLogCollector.makeLatestCollector();
            for (int i = 0; i < aCase; i++) {
                collector.fold((HyperLogLogCollector) complexColumn.getRowValue(i));
            }
            Assert.assertEquals(baseCollector.estimateCardinality(), collector.estimateCardinality(), 0.0);
        }
    }
}

Example 3 with SmooshedFileMapper

use of org.apache.druid.java.util.common.io.smoosh.SmooshedFileMapper in project druid by druid-io.

the class CompressedVSizeColumnarIntsSerializerTest method checkV2SerializedSizeAndData.

private void checkV2SerializedSizeAndData(int chunkSize) throws Exception {
    File tmpDirectory = temporaryFolder.newFolder();
    FileSmoosher smoosher = new FileSmoosher(tmpDirectory);
    final String columnName = "test";
    GenericIndexedWriter genericIndexed = GenericIndexedWriter.ofCompressedByteBuffers(segmentWriteOutMedium, "test", compressionStrategy, Long.BYTES * 10000);
    CompressedVSizeColumnarIntsSerializer writer = new CompressedVSizeColumnarIntsSerializer(columnName, segmentWriteOutMedium, vals.length > 0 ? Ints.max(vals) : 0, chunkSize, byteOrder, compressionStrategy, genericIndexed);
    writer.open();
    for (int val : vals) {
        writer.addValue(val);
    }
    final SmooshedWriter channel = smoosher.addWithSmooshedWriter("test", writer.getSerializedSize());
    writer.writeTo(channel, smoosher);
    channel.close();
    smoosher.close();
    SmooshedFileMapper mapper = Smoosh.map(tmpDirectory);
    CompressedVSizeColumnarIntsSupplier supplierFromByteBuffer = CompressedVSizeColumnarIntsSupplier.fromByteBuffer(mapper.mapFile("test"), byteOrder);
    ColumnarInts columnarInts = supplierFromByteBuffer.get();
    for (int i = 0; i < vals.length; ++i) {
        Assert.assertEquals(vals[i], columnarInts.get(i));
    }
    CloseableUtils.closeAll(columnarInts, mapper);
}

Example 4 with SmooshedFileMapper

use of org.apache.druid.java.util.common.io.smoosh.SmooshedFileMapper in project druid by druid-io.

the class V3CompressedVSizeColumnarMultiIntsSerializerTest method checkV2SerializedSizeAndData.

private void checkV2SerializedSizeAndData(int offsetChunkFactor, int valueChunkFactor) throws Exception {
    File tmpDirectory = FileUtils.createTempDir(StringUtils.format("CompressedVSizeIndexedV3WriterTest_%d_%d", offsetChunkFactor, offsetChunkFactor));
    FileSmoosher smoosher = new FileSmoosher(tmpDirectory);
    int maxValue = vals.size() > 0 ? getMaxValue(vals) : 0;
    try (SegmentWriteOutMedium segmentWriteOutMedium = new OffHeapMemorySegmentWriteOutMedium()) {
        CompressedColumnarIntsSerializer offsetWriter = new CompressedColumnarIntsSerializer(TEST_COLUMN_NAME, segmentWriteOutMedium, offsetChunkFactor, byteOrder, compressionStrategy, GenericIndexedWriter.ofCompressedByteBuffers(segmentWriteOutMedium, "offset", compressionStrategy, Long.BYTES * 250000));
        GenericIndexedWriter genericIndexed = GenericIndexedWriter.ofCompressedByteBuffers(segmentWriteOutMedium, "value", compressionStrategy, Long.BYTES * 250000);
        CompressedVSizeColumnarIntsSerializer valueWriter = new CompressedVSizeColumnarIntsSerializer(TEST_COLUMN_NAME, segmentWriteOutMedium, maxValue, valueChunkFactor, byteOrder, compressionStrategy, genericIndexed);
        V3CompressedVSizeColumnarMultiIntsSerializer writer = new V3CompressedVSizeColumnarMultiIntsSerializer(TEST_COLUMN_NAME, offsetWriter, valueWriter);
        writer.open();
        for (int[] val : vals) {
            writer.addValues(new ArrayBasedIndexedInts(val));
        }
        final SmooshedWriter channel = smoosher.addWithSmooshedWriter("test", writer.getSerializedSize());
        writer.writeTo(channel, smoosher);
        channel.close();
        smoosher.close();
        SmooshedFileMapper mapper = Smoosh.map(tmpDirectory);
        V3CompressedVSizeColumnarMultiIntsSupplier supplierFromByteBuffer = V3CompressedVSizeColumnarMultiIntsSupplier.fromByteBuffer(mapper.mapFile("test"), byteOrder);
        ColumnarMultiInts columnarMultiInts = supplierFromByteBuffer.get();
        Assert.assertEquals(columnarMultiInts.size(), vals.size());
        for (int i = 0; i < vals.size(); ++i) {
            IndexedInts subVals = columnarMultiInts.get(i);
            Assert.assertEquals(subVals.size(), vals.get(i).length);
            for (int j = 0, size = subVals.size(); j < size; ++j) {
                Assert.assertEquals(subVals.get(j), vals.get(i)[j]);
            }
        }
        CloseableUtils.closeAll(columnarMultiInts, mapper);
    }
}

Example 5 with SmooshedFileMapper

use of org.apache.druid.java.util.common.io.smoosh.SmooshedFileMapper in project druid by druid-io.

the class CompressedColumnarIntsSerializerTest method checkV2SerializedSizeAndData.

private void checkV2SerializedSizeAndData(int chunkFactor) throws Exception {
    File tmpDirectory = FileUtils.createTempDir(StringUtils.format("CompressedIntsIndexedWriterTest_%d", chunkFactor));
    FileSmoosher smoosher = new FileSmoosher(tmpDirectory);
    CompressedColumnarIntsSerializer writer = new CompressedColumnarIntsSerializer("test", segmentWriteOutMedium, chunkFactor, byteOrder, compressionStrategy, GenericIndexedWriter.ofCompressedByteBuffers(segmentWriteOutMedium, "test", compressionStrategy, Long.BYTES * 10000));
    writer.open();
    for (int val : vals) {
        writer.addValue(val);
    }
    final SmooshedWriter channel = smoosher.addWithSmooshedWriter("test", writer.getSerializedSize());
    writer.writeTo(channel, smoosher);
    channel.close();
    smoosher.close();
    SmooshedFileMapper mapper = Smoosh.map(tmpDirectory);
    // read from ByteBuffer and check values
    CompressedColumnarIntsSupplier supplierFromByteBuffer = CompressedColumnarIntsSupplier.fromByteBuffer(mapper.mapFile("test"), byteOrder);
    ColumnarInts columnarInts = supplierFromByteBuffer.get();
    Assert.assertEquals(vals.length, columnarInts.size());
    for (int i = 0; i < vals.length; ++i) {
        Assert.assertEquals(vals[i], columnarInts.get(i));
    }
    CloseableUtils.closeAndWrapExceptions(columnarInts);
    mapper.close();
}